Nondictatorial & Nonmanipulable axioms (was Re: York's

[Steve Eppley]

Mike O. wrote:

>Since no method is un-manipulable, as Gibbard & Satterthwaite
>use the term in their mysterious, un-posted theorem, what's the
>point for Bruce to say in his May 2 letter that Condorcet's
>method isn't un-manipulable by that un-stated definition?

Bruce said that in reply to one of my questions.  I don't think his 
intention was to single out Condorcet as failing that criterion.

I knew that Condorcet, like other methods, must fail at least one of
Arrow's inconsistent axioms and at least one of Gibbard-Satterthwaite's
inconsistent axioms.  I had asked which ones it fails, since Mike 
Saari had stated that Condorcet (like other ranked ballot systems)
fails Arrow's nondictatorial axiom (and that axiom is also one of 
Gibbard-Satterthwaite's axioms).  Bruce's answer is that Mike Saari 
is incorrect.

You're right; manipulability isn't necessarily a problem.  The flip
side of it is that voters' choices can make a difference--something
we expect in a democracy. 

--Steve